Application ScenariosA body-in-white spot-welding cell running four IRB 7600-325 (325 kg payload) robots was logging intermittent “DC Overvoltage 32014” faults on Robot #2 during the gun-retract portion of the weld cycle — the gun closes (motor motoring), then the servo releases and the arm snaps back (motor generating), pushing ~4–6 kW regen into the DC link for 120–180 ms. The cell’s weld schedule is 42 spots per panel, 18 seconds per panel, so the fault showed up 6–10× per shift, each time forcing a 45-second drive reset and breaking the line takt. The maintenance team scoped the DC link on Robot #2’s IRC5 drive cabinet: the 3HAC14551-2​ C3 unit (one of three capacitor modules in that Size-4 dual-drive stack — C1 + C2 + C3 in parallel) showed ESR climb from the original < 12 mΩ to ~31 mΩ on Ch-C3 (measured via the drive’s internal power-diagnostics trend — ABB’s DriveWare can log DC-link ripple RMS). The C3’s ability to sink the regen pulse had degraded after 9 years of 24/7 two-shift running, so the bus peaked at 412 V DC during retract (trip threshold = 410 V on that drive size). The swap: power down the IRC5 cabinet, LOTO, wait 7 minutes for the C3’s built-in discharge resistor to bleed the bus below 50 V, undo the M8 busbar bolts + the M4 fastening screws + the temperature/balance plug, lift the 3HAC14551-2​ out (2.48 kg, manageable one-handed), seat the new one, torque M8 to 8 N·m + M4 to 1.2 N·m, reconnect the plug, power up, auto-comm the drive. Faults gone across a 14-month follow. The cell lead: “The 3HAC14551-2​ isn’t sexy, but when ESR drifts on a 325 kg robot doing 42 spots per panel, it’s the difference between takt and reset.” This case frames the C3’s role: it’s the sacrificial energy sponge in the IRC5 drive stack, and its electrolytic wear is the silent fault that shows up as “sporadic DC OV” before becoming “IGBT blow.”

Parameter

Note on C-value conflict in public sources: Some distributors list the 3HAC14551-2​ as “1200 µF / 400 V DC” , others as “3 × 4700 µF = 14,100 µF / 450 V DC” . The discrepancy is real — ABB’s C1/C2/C3 stack varies by drive Size (Size 3 may carry one C-unit ~1200–4700 µF range; Size 4 dual-drive stacks C1+C2+C3 in parallel, each C-unit 3×4700 µF). The 3HAC14551-2​ specifically is the C3 position; if your drive BOM (3HAC020536-014) calls for C3 = 3HAC14551-2, assume the 14.1 mF / 450 V config unless your drive nameplate says otherwise. Verify against your IRC5 Drive Module Size before ordering.Technical Principles and Innovative Values

• Innovation Point 1: 3-Can Parallel Bank with < 15 mΩ ESR.​ The 3HAC14551-2​ C3 isn’t one big capacitor — it’s three 4700 µF electrolytic cans paralleled on a shared copper-busbar bracket. Paralleling cuts ESR: three cans at ~40 mΩ each → ~13 mΩ combined, which means lower self-heating under the 20–30 A RMS ripple that a 325 kg robot’s wrist-axes deceleration throws at the DC link. Lower ESR = lower ΔT = longer life. A single 14 mF can at the same ripple would run 15–20°C hotter — the C3’s 3-can layout is a thermal design choice, not just a “more µF” choice.
• Innovation Point 2: Bolt-On Busbar Integration with Balance/Temp Plug.​ Unlike DIN-mount capacitor modules (which use screw terminals), the 3HAC14551-2​ lands directly on the IRC5 drive’s DC busbar with M8 copper bolts — same busbar that feeds the IGBT module. This minimizes parasitic inductance between the C3 and the IGBT (critical: during the IGBT’s turn-off, dI/dt ~ thousands of A/µs, and L·dI/dt adds to Vce overshoot — shorter busbar path = less L). The small 2-pin plug on the C3 carries a temperature-sense NTC and a can-balance tap back to the Drive Module’s power-board — if one can drifts or a busbar bolt loosens, the drive logs “Capacitor Warning” before it becomes “IGBT fault.”
• Innovation Point 3: Built-In Discharge Resistor, But Not a Substitute for LOTO Wait.​ The 3HAC14551-2​ has a permanent discharge resistor across the busbar lugs (value chosen so 400 V → < 60 V in ~5–7 min). This is why ABB’s safety manual says “wait 5 min after power-off before touching DC link.” But — and this matters — the discharge resistor is sized for normal conditions. If the C3’s internal connection has a partial open (aged solder joint), the resistor may not discharge that one can fully. The IRC5 Drive Module’s pre-charge circuit also checks “DC link voltage < 50 V” before closing contactor — if the C3 is leaky, the pre-charge may fault “DC Link Uv” on next power-up. Point: the discharge resistor is a convenience, not a guarantee — always verify with a CAT III meter.
• Innovation Point 4: C1/C2/C3 Stack Architecture.​ In a Size-4 IRC5 dual-drive (e.g., IRB 7600 + positioner shared drive), the DC link uses C1 + C2 + C3 in parallel — three 3HAC14551-2-class modules (C1/C2 may be different order codes — 3HAC14549-2 / 3HAC14550-2 — but mechanically identical footprint). This distributed-cap approach means one C-unit can fail (ESR drift, can leak) and the other two still carry the bus — the drive derates current but doesn’t trip. That’s why “sporadic DC OV on Robot #2 only during retract” is the classic C3-failing signature: the other two C-units are fine, so steady-state is OK, but the regen pulse peaks the bus because the weak C3 can’t sink the ripple current.

Application Cases and Industry ValueCase 1 – Automotive Press Shop, IRB 7600-325 Spot Welding (the scenario above, expanded). Four 7600s on a 42-spot-per-panel cadence, two-shift, 280 days/year. The #2 robot’s C3 (9 years old) ESR climbed to 31 mΩ, triggering “DC OV 32014” 6–10×/shift. The swap took one tool-change window (22 minutes: 8 min power-down+LOTO+discharge wait, 6 min unbolt C3, 4 min seat new, 2 min torque, 2 min plug, power-up + DriveWare auto-comm). Post-swap, DC-link ripple RMS during retract dropped from 18 V pp to 6 V pp, and the 32014 fault never returned across 14 months. The plant reliability engineer tracked it: “Each 32014 fault cost ~0.7 panel (weld-incomplete reject) + 45 sec reset = ~€180/fault. We were at ~8 faults/shift × 2 shifts = 16/day × 280 days = 4480 faults/year. Swapping the 3HAC14551-2​ C3 eliminated ~85% of those (the rest were gun-sticking, not DC OV). ROI = one C3 (~€900) vs. ~€800K/year in avoided rejects. Math isn’t close.”Case 2 – Palletizing Cell, IRB 6600-250, Beverage Plant. Single IRB 6600 on a 14-case/sec palletizer, IRC5 Single cabinet, Size-3 drive (one C3 only, no C1/C2 — Size 3 is single-drive, C3 is the only DC-link cap). After 11 years, the C3’s discharge resistor failed open (one of the two parallel legs on the resistor cracked — thermal cycling from the cabinet’s +48°C ambient near the VFD lineup). Symptom: after a weekend E-stop test, the cabinet wouldn’t re-power — pre-charge contactor clicked, then “DC Link Uv” + “Precharge Timeout.” The tech measured DC link 18 hours after power-off: 127 V DC still on the bus (should be < 60 V after 7 min). The 3HAC14551-2​ C3’s discharge resistor had one leg open, so only half the resistance was bleeding — T = RC doubled. Swapped C3, cabinet powered normally. The maintenance lead: “We’d never have guessed ‘discharge resistor’ from ‘precharge timeout.’ The 3HAC14551-2​ was the root. Now we check DC-link voltage 10 min after power-off during annual PM — if it’s > 60 V, the C3’s discharge is suspect.”